13 papers
Measuring Empowerment from Offline Data for Intrinsically Motivated Exploration
Ishir Garg, Vivek Myers, Qiyang Li +3
Large, unlabeled datasets of prior experience have become increasingly available for agentic tasks in domains such as robotics and coding. A common recipe for these tasks is to tra…
Scaling Empowered Agents
Catherine Ji, Vivek Myers, Joshua Lin +3
Empowerment measures the capacity of an agent to actively control its environment. In this work, we explain the connection between skill learning and empowerment. While skill learn…
The Geometry of Empowerment
Catherine Ji, Vivek Myers, Sergey Levine +1
Empowerment captures the capacity for an agent to actively control its environment. While conceptually appealing as an information-theoretic quantity, the connection between empowe…
Multistep Quasimetric Learning for Scalable Goal-conditioned Reinforcement Learning
Bill Chunyuan Zheng, Vivek Myers, Benjamin Eysenbach +1
Learning how to reach goals in an environment is a longstanding challenge in AI, yet reasoning over long horizons remains a challenge for modern methods. The key question is how to…
Training LLM Agents to Empower Humans
Evan Ellis, Vivek Myers, Jens Tuyls +3
Assistive agents should not only take actions on behalf of a human, but also step out of the way and cede control when there are important decisions to be made. However, current me…
Offline Goal-conditioned Reinforcement Learning with Quasimetric Representations
Vivek Myers, Bill Chunyuan Zheng, Benjamin Eysenbach +1
Approaches for goal-conditioned reinforcement learning (GCRL) often use learned state representations to extract goal-reaching policies. Two frameworks for representation structure…
Temporal Representation Alignment: Successor Features Enable Emergent Compositionality in Robot Instruction Following
Vivek Myers, Bill Chunyuan Zheng, Anca Dragan +2
Effective task representations should facilitate compositionality, such that after learning a variety of basic tasks, an agent can perform compound tasks consisting of multiple ste…
Horizon Generalization in Reinforcement Learning
Vivek Myers, Catherine Ji, Benjamin Eysenbach
We study goal-conditioned RL through the lens of generalization, but not in the traditional sense of random augmentations and domain randomization. Rather, we aim to learn goal-dir…
Learning to Assist Humans without Inferring Rewards
Vivek Myers, Evan Ellis, Sergey Levine +2
Assistive agents should make humans' lives easier. Classically, such assistance is studied through the lens of inverse reinforcement learning, where an assistive agent (e.g., a cha…
Policy Adaptation via Language Optimization: Decomposing Tasks for Few-Shot Imitation
Vivek Myers, Bill Chunyuan Zheng, Oier Mees +2
Learned language-conditioned robot policies often struggle to effectively adapt to new real-world tasks even when pre-trained across a diverse set of instructions. We propose a nov…
Accelerating Goal-Conditioned RL Algorithms and Research
Michał Bortkiewicz, Władysław Pałucki, Vivek Myers +4
Self-supervision has the potential to transform reinforcement learning (RL), paralleling the breakthroughs it has enabled in other areas of machine learning. While self-supervised…
Learning Temporal Distances: Contrastive Successor Features Can Provide a Metric Structure for Decision-Making
Vivek Myers, Chongyi Zheng, Anca Dragan +2
Temporal distances lie at the heart of many algorithms for planning, control, and reinforcement learning that involve reaching goals, allowing one to estimate the transit time betw…
Inference via Interpolation: Contrastive Representations Provably Enable Planning and Inference
Benjamin Eysenbach, Vivek Myers, Ruslan Salakhutdinov +1
Given time series data, how can we answer questions like "what will happen in the future?" and "how did we get here?" These sorts of probabilistic inference questions are challengi…